Sabah Nushra

SL-7: A controlled multimodal benchmark for diagnosing structured reasoning failures in vision-language models using prerequisite-aware evaluation.

BILAL-Bench is a Bengali benchmark for evaluating fabrication and source attribution in open language models across Qur’an, Hadith, Fiqh, contemporary Islamic issues, and adversarial false-premise prompts.

Restore and colorize old black and white photos, a classical restoration pipeline feeding transfer-learned deep colorization models. Live demo on 🤗 Spaces.

Depth and material-resolved comparison of VNIR hyperspectral imaging and EM61 electromagnetic induction for buried-target detection.

TConnect is a prescription management website, where doctors can see a patient's medical documents if the patient gives permission.

A hostel management mobile app. Here a student can book a room, give complaints to the authority, and can see notices.